Molecular Psychiatry (2020) 25:1215–1228https://doi.org/10.1038/s41380-019-0384-6

ARTICLE

Hippocampal mossy cell involvement in behavioral and neurogenicresponses to chronic antidepressant treatment

Seo-Jin Oh1● Jia Cheng2

● Jin-Hyeok Jang1● Jeffrey Arace2 ● Minseok Jeong1

● Chang-Hoon Shin1 ● Jeongrak Park1 ●

Junghee Jin2● Paul Greengard2

● Yong-Seok Oh 1

Received: 5 April 2018 / Revised: 29 January 2019 / Accepted: 14 February 2019 / Published online: 5 March 2019© Springer Nature Limited 2019

AbstractMost antidepressants, including selective serotonin reuptake inhibitors (SSRIs), initiate their drug actions by rapid elevationof serotonin, but they take several weeks to achieve therapeutic onset. This therapeutic delay suggests slow adaptive changesin multiple neuronal subtypes and their neural circuits over prolonged periods of drug treatment. Mossy cells are excitatoryneurons in the dentate hilus that regulate dentate gyrus activity and function. Here we show that neuronal activity ofhippocampal mossy cells is enhanced by chronic, but not acute, SSRI administration. Behavioral and neurogenic effects ofchronic treatment with the SSRI, fluoxetine, are abolished by mossy cell-specific knockout of p11 or Smarca3 or by aninhibition of the p11/AnxA2/SMARCA3 heterohexamer, an SSRI-inducible protein complex. Furthermore, simplechemogenetic activation of mossy cells using Gq-DREADD is sufficient to elevate the proliferation and survival of theneural stem cells. Conversely, acute chemogenetic inhibition of mossy cells using Gi-DREADD impairs behavioral andneurogenic responses to chronic administration of SSRI. The present data establish that mossy cells play a crucial role inmediating the effects of chronic antidepressant medication. Our results indicate that compounds that target mossy cellactivity would be attractive candidates for the development of new antidepressant medications.

Introduction

Major depressive disorder (MDD) is the most prevalentmental illness, of which lifetime prevalence is estimated tobe as high as 16.2% in the United States. This diseaseencompasses various symptoms, including anhedonia,depressed mood, increased stress sensitivity, helplessness,apathy, shift towards negative emotions (sadness, emotionalnumbness, irritability, and anxiety), and cognitive deficits

[1, 2]. Selective serotonin reuptake inhibitors (SSRIs) arethe most widely used class of antidepressants [3], but SSRIsgenerally take several weeks to show therapeutic effects,despite their immediate effect on serotonin neurotransmis-sion [3, 4]. This therapeutic delay suggests the existence ofslow adaptive changes in neural circuits over a long-lastingperiod of drug treatment, possibly involving changes ingene expression and protein translation [5–7]. In addition,although significant progress has been made in treatingdepression by SSRIs, only a little more than half of severelydepressed patients has been found to respond to this class ofdrugs [8]. Our knowledge of the molecular mechanismsunderlying therapeutic responses to long-term treatmentwith SSRIs, as well as the side effects of the drugs, have yetto be established at the level of neuronal cell types.

Mossy cells reside in the hilus of the dentate gyrus andmake synaptic connections with various types of cells, such asgranule cells and basket cells [9–13]. Moreover, they arepositioned to integrate numerous inputs mainly from multiplegranule cells, CA3c pyramidal cells, as well as from the localGABAergic interneurons including basket cells and hilarperforant path-associated (HIPP) cells and then to propagate asignal over the longitudinal axis of the hippocampus [13–15].

These authors contributed equally: Seo-Jin Oh, Jia Cheng

* Yong-Seok Ohysoh2040@dgist.ac.kr

1 Department of Brain-Cognitive Science, Daegu-GyeongbukInstitute of Science and Technology (DGIST), Hyenpung-myeon, Dalseong-gun, Daegu, Republic of Korea

2 Laboratory of Molecular and Cellular Neuroscience, TheRockefeller University, New York, NY 10065, USA

Supplementary information The online version of this article (https://doi.org/10.1038/s41380-019-0384-6) contains supplementarymaterial, which is available to authorized users.

1234

5678

90();,:

1234567890();,:

Of particular interest, recent studies have shown thatmossy cells are highly enriched in selective subsets ofmonoamine receptors, proving mossy cells to be immediateregulatory targets of monoaminergic inputs to the hippo-campus [12, 16, 17]. They also highly express type IIglucocorticoid receptors, which sense the stress hormone,cortisol [18]. Mossy cells fire frequently in multiple placefields in various environments, in contrast to granulecells which fire sparsely [19–21]. The neuronal activity ofmossy cells within the dentate gyrus circuitry is crucial forthe proper function of the hippocampus [22]. Indeed,degeneration of mossy cells results in the dysregulation ofgranule cell excitability, which leads to abnormal behaviorssuch as elevated anxiety, and impaired pattern separation[23]. Despite the potential implication of the mossy cellsin the affective disorders, as well as in the actions ofmonoaminergic drugs, a detailed analysis of the role ofmossy cells in antidepressant responses has not yet beencarried out.

p11 (S100A10) is a key factor involved in regulatingaffective state and in determining responses to anti-depressants [24, 25]. It is downregulated in depressedhumans and rodents and is induced by chronic but notby acute administration of antidepressants in rodents[24, 26, 27]. Global abolition of p11 in mice produces adepressive state [24]. Removal of p11 from specific classesof p11-enriched neurons produces a variety of depressiveeffects, depending on the subtype of p11-containing neu-rons that is targeted [28–37].

p11 was initially identified within a heterotetramericcomplex where it forms with Annexin A2 (AnxA2) [38],and this complex form of p11 and AnxA2 can be induced inthe hippocampal dentate gyrus by chronic treatment with anSSRI. Previous studies identified SMARCA3 as a bindingpartner of the p11/AnxA2 complex and both p11 andSMARCA3 are crucial for behavioral and neurogenicresponses to chronic antidepressant administration [34].Chronic SSRI administration increased the level of theheterohexameric complex of p11/AnxA2/SMARCA3 in thehippocampus [34]. In addition, p11 and SMARCA3are highly enriched in hilar mossy cells and basket cellsin the hippocampal dentate gyrus [34]. However, a celltype-specific role of the p11/AnxA2/SMARCA3 hetero-hexameric complex in antidepressant action has not yetbeen explored. Here we examine the role of mossy cells inmediating the actions of the antidepressant, fluoxetine.Results from this study have suggested that the hetero-hexameric complex of p11/AnxA2/SMARCA3 regulatesthe activity of mossy cells, which is required for behavioraland neurogenic responses to chronic antidepressantmedication.

Materials and methods

Animal breeding

All procedures involving animals were approved by theAnimal Care and Use Committee of the Daegu GyeongbukInstitute of Science & Technology (DGIST, Korea) and bythe Rockefeller University (USA) Institutional Animal Careand Use Committee. Transgenic mouse lines were used inthis study: MC-Cre ([Calcrl]-Cre, JAX stock #023014),D2-Cre ([Drd2]-Cre, GENSAT, Clone #ER44), tdTomatoReporter (JAX stock #007908), p11 cKO (p11(f/f);D2-Cre),Smarca3 cKO(Smarca3(f/f);D2-Cre, Smarca3(f/f);MC-Cre). Allthe animal works were done at the laboratory animalresource center (LARC) of DGIST or the comparativebioscience center (CBC) of the Rockefeller University. Themouse breeding methods and housing conditions aredescribed in Supplementary Materials and Methods.

Drug treatments

Details of drug treatments were included in SupplementaryMaterials and Methods.

Plasmid constructions and immunoprecipitation

All plasmid constructions and immunoprecipitation wereperformed as previously described [34]. See detailedmethods in Supplementary Materials and Methods.

Stereotaxic surgery

Mice were anesthetized by intraperitoneal injection of avertin(250mg/kg), and stereotaxic injection of AAV constructs wasperformed into the hippocampal hilus regions. (coordinatesfor dorsal: AP −2.1mm, ML ±1.4mm, DV −1.95mm,ventral: AP −3.3mm, ML ±2.7mm, DV −3.6 mm). Seedetailed methods in Supplementary Materials and Methods.

Behavior assessments

Mood and anxiety-related behaviors (novelty suppressedfeeding [NSF] test, tail suspension test [TST], elevated plusmaze [EPM], and light/dark box [LDB] test) and locomotoractivity (open field [OF] test) were tested as describedin Supplementary Materials and Methods.

Chronic unpredictable mild stress paradigm

This paradigm was described in detail in SupplementaryMaterials and Methods.

1216 S.-J. Oh et al.

Immunohistochemistry

All mice were perfused transcardially with PBS, followedby 4% paraformaldehyde (PFA) in PBS and post fixed inthe same solution overnight at 4 °C. The brains were cor-onally cut into 40-μm-thick sections with a Cryostat(CM3050S, Leica). Sections were processed using a free-floating procedure. Detailed description of antibody pre-paration, antigen retrieval, image acquisition, and quantifi-cation is provided in Supplementary Materials andMethods.

BrdU labeling and neurogenesis assay

For the neurogenesis assay, the mice were labeled withBrdU solution (200 mg/kg) for 3 h prior to sacrifice byperfusion. See detailed methods in Supplementary Materialsand Methods.

Electrophysiological recordings of mossy cells

Slice preparation and electrophysiology were performed asdescribed previously [39]. Experimental details are includedin Supplementary Materials and Methods.

Data analysis and statistics

All data were presented as means ± SEM. Statistical ana-lysis was conducted using GraphPad Prism Version 7.0a.Two group comparisons were done by two-tailed, unpairedor paired Student’s t-test. Multiple group comparisons wereassessed using a one-way or two-way ANOVA, followedby the appropriate post hoc test. Significance thresholdswere as follows: *p < 0.05, **p < 0.01, ***p < 0.001.

Results

Effects of genetic deletion of p11 or Smarca3 inhippocampal mossy cells on behavioral responses tochronic SSRI administration

Previous studies identified SMARCA3 as a binding partnerof the p11/AnxA2 complex and, using constitutive Smarca3KO mice, demonstrated its crucial role for neurogenic andbehavioral responses to chronic antidepressant administra-tion. Those studies also showed that p11 and SMARCA3were highly enriched in hippocampal mossy cells [34].However, a possible mossy cell-specific involvement of thep11/AnxA2/SMARCA3 complex in antidepressant actionswas not studied. Therefore, we generated mossy cell-specific deletion of p11 or Smarca3 conditional KO mice toidentify the cell type-specific role of the ternary complex.

To investigate the possibility of a mossy cell-specific role ofp11 and/or Smarca3 in the actions of antidepressants, weused a transgenic Cre driver line to target hippocampalmossy cells. According to previous reports, dopamine D2receptor gene expression is limited to mossy cells in thedentate gyrus [40, 41]. We validated dopamine D2 receptorpromoter driven Cre recombinase expression in a mossycell-specific manner in the dentate gyrus by crossing a D2-Cre ([Drd2]-Cre) driver mouse line with a tdTomatoreporter line (Fig. 1a, b). Then, we crossed them with a p11or Smarca3 floxed conditional line to achieve specificdeletion of p11 or Smarca3 in mossy cells (Fig. 1c, f,Figures S1a and S1b). After chronic administration of SSRIto a p11 or Smarca3 conditional KO (cKO) line, we carriedout the novelty suppressed feeding test to investigatedepression-like states. We found that the latency toapproach food pellet was reduced in the control mice(p11(f/f) and Smarca3(f/f)), but not in p11 cKO (p11(f/f);D2-Cre)or Smarca3 cKO (Smarca3(f/f);D2-Cre) mice (Fig. 1d, g),but that their home cage feeding level was unaffected(Fig. 1e, h). We further validated the effect of geneticdeletion of Smarca3 in the mossy cells using another typeof mossy cells-specific Cre mouse line, MC-Cre ([Calcrl]-Cre) (Fig. 1i, j). This MC-Cre line was very specific in thehippocampus, but exhibited wide activity in both hilarmossy cells of dentate gyrus and some pyramidal cells ofthe CA3 region [23]. We deleted the Smarca3 gene bycrossing MC-Cre mice with Smarca3 floxed conditionalmice (Smarca3(f/f)) (Fig. 1k) and conducted the noveltysuppressed feeding test (NSF) and the tail-suspension test(TST) after chronic administration of SSRI. We foundthat the latency to feed in the NSF and the immobilityin the TST were reduced in the control mice byfluoxetine treatment (Smarca3(f/f)), but not in Smarca3 cKO(Smarca3(f/f);MC-Cre) (Fig. 1l, m, Figure S2a), without anysignificant change in locomotor activity (Figure S2b).

All of these data indicate that selective deletion of p11 orSmarca3 in dentate mossy cells affects the antidepressantresponses. These results suggest that p11 and Smarca3 inmossy cells are involved in mediating the behavioralresponses to chronic antidepressant treatment.

Effect of mossy cell-specific inhibition of the p11/AnxA2/SMARCA3 complex on neurogenic andbehavioral responses to chronic antidepressanttreatment

Although we evaluated the effects of genetic deletion of p11or Smarca3 gene in mossy cells on antidepressant responsesusing cell type-specific KO mice, these transgenic approa-ches could not rule out the possibility that these behavioralchanges are due to developmental defects or off-targetdeletion of p11 or Smarca3 gene in those cKO mice. Thus,

Hippocampal mossy cell involvement in behavioral and neurogenic responses to chronic antidepressant. . . 1217

we further examined the neurogenic and behavioral effectsof mossy cell-specific inhibition of the p11/AnxA2/SMARCA3 complex in adult mice. To this end, wedeveloped recombinant constructs to inhibit the assembly ofthe p11/AnxA2/SMARCA3 complex. Our previous studyshowed that the p11/AnxA2 heterotetramer formed a stableheterohexameric complex with either SMARCA3 orAHNAK1, using very similar recognition principles. Theshort peptide derived from the binding region of

SMARCA3 or AHNAK1 makes a stable complex with thep11/AnxA2 heterotetramer by occupying the hydrophobicbinding pocket created on the surface of the P11/AnxA2heterotetrameric complex [34]. Based on information aboutthe co-crystalization structure of the ternary complex, wedesigned a recombinant inhibitor of the p11/AnxA2/SMARCA3 complex (P11/AnxA2/SMARCA3 complexinhibitory peptide-AcGFP1 fusion construct (PASIP-AcGFP1)) along with an inert control (control AcGFP1)

1218 S.-J. Oh et al.

(Fig. 2a). We assumed that this recombinant inhibitor pro-tein competes with endogenous SMARCA3 on the bindingpocket and thus blocks the functional assembly of the activeternary complex, the p11/AnxA2/SMARCA3 hetero-hexamer (Fig. 2b). By conducting an in vitro co-immunoprecipitation assay, we confirmed that the recom-binant PASIP-AcGFP1 construct was able to successfullyblock the assembly of the P11/AnxA2/SMARCA3 complex(Fig. 2c). We further generated Cre-dependent AAVs todeliver those PASIP-AcGFP1 constructs into the hippo-campal mossy cells. Cre-dependent AAVs were injectedlocally into hilus regions of MC-Cre transgenic mice(Fig. 2d, Figures S3a and S3b). Because MC-Cre mice wereknown to display Cre-recombination in both dentate mossycells and a minor pyramidal subpopulation in the CA3cregion [23], we delivered recombinant AAV solution in themiddle of the hilus region to avoid the unnecessary diffu-sion of viral particles. Our histological analysis confirmedthat AcGFP1 signal was found exclusively in the hilus and

the inner molecular layer (IML), where somas and axonalfibers of mossy cells localize, respectively, but not in theCA3 region where pyramidal cells localize (Figure S3c),ensuring the hippocampal subregion specificity. In addition,we confirmed that AcGFP1-fusion proteins (controlAcGFP1, PASIP-AcGFP1) were specifically expressed inmossy cells, but not in other GABAergic interneurons,including PV+ basket cells and NPY+HIPP cells (Fig. 2e,Figure S3d), ensuring the cell type specificity in the dentategyrus.

After preventing formation of the p11/AnxA2/SMARCA3 complex with mossy cell-specific expression ofthe PASIP-AcGFP1 construct, we examined the neurogenicand behavioral outcome of chronic administration of SSRI(Fig. 2f). It is well established that adult neurogenesis in thedentate gyrus is induced in response to chronic anti-depressant treatment [42–44]. In control and PASIP-AcGFP1 mice, we assessed chronic fluoxetine-inducedproliferation activity of neural progenitors. A significantincrease in BrdU+ proliferating cells was observed in thesubgranular zone (SGZ) of control AcGFP1 mice afterchronic fluoxetine treatment, which was abolished in theSGZ of PASIP-AcGFP1 mice (p= 0.73) (Fig. 2g, h, Fig-ures S4a and S4b). In addition, we analyzed the expressionlevel of doublecortin (DCX), which represent a snapshot ofnewborn postmitotic cells undergoing neuronal differentia-tion and maturation [45]. Chronic fluoxetine administrationincreased the DCX immunofluorescence signal in the SGZof control AcGFP1 mice, but the effect of fluoxetine wasattenuated in the SGZ of PASIP-AcGFP1 mice (Fig. 2i, j,Figures S4c and S4d). We found that the inhibitory effect ofPASIP-AcGFP1 expression on fluoxetine-induced neuro-genic activities was observed consistently along the rostro-caudal axis of the dentate gyrus (Figures S4b and S4d).

We next investigated the possible functional significanceof the p11/AnxA2/SMARCA3 complex in mossy cells forbasal locomotor activity, anxiety-related behaviors andSSRI-induced behavioral changes. Control AcGFP1 andPASIP-AcGFP1 mice did not display a difference in basallocomotor activity when monitored using the OF test(Figure S5a). In addition, inhibition of the p11/AnxA2/SMARCA3 complex failed to show a consistent effect onmultiple anxiety-related behaviors. Control AcGFP1 andPASIP-AcGFP1 mice showed a difference in the light/darkbox [LDB] test (p= 0.0064) (Figure S5c), but not in theother two anxiety-related behaviors (thigmotaxis test (p=0.74), Figure S5b; the elevated plus maze test [EPM] (p=0.587), Figure S5d). In addition, they did not display anybaseline difference in depression-like behavioral tests (tailsuspension test [TST], Figure S5e; novelty-suppressedfeeding [NSF] test, Fig. 2k). However, the latency to feedwas decreased after chronic fluoxetine administration incontrol AcGFP1 mice, but not in PASIP-AcGFP1 mice

Fig. 1 Mossy cell-specific knockout of p11 or Smarca3 gene abolishesbehavioral responses to chronic SSRI treatment. a Schematic design toconfirm mossy cell-specific Cre recombination in the hippocampus ofD2-Cre transgenic mice. b Co-localization of Cre-dependent reporter(tdTomato, red) with CRT, a mossy cell marker (green, filled arrow-heads), but not with PV, a PV+ basket cell marker (green, openarrowheads). Scale bar, 25 µm. c Schematic illustration of geneticdeletion of p11 gene in mossy cells using a D2-Cre line. d Noveltysuppressed feeding (NSF) test. Bar graphs showing latency to foodpellet. Control and p11 cKO group have genotype of p11(f/f) withoutCre, and p11(f/f) with D2-Cre, respectively. Two-way ANOVA, [gen-otype × drug interaction F(1,54)= 5.031; p= 0.0290, genotype factorF(1,54)= 7.435; p= 0.0086, drug factor F(1,54)= 6.728; p=0.0122], followed by the Turkey’s post hoc test. e Assessment of thehunger level in control and p11 cKO group using home cage feedingtest right after NSF test of the same set of animals. f Schematicillustration of genetic deletion of Smarca3 gene in mossy cells using aD2-Cre line. g NSF tests. Control and Smarca3 cKO group havegenotype of Smarca3(f/f) without Cre, and Smarca3(f/f) with D2-Cre,respectively. Two-way ANOVA, [genotype × drug interaction F(1,44)= 9.111; p= 0.0042, genotype factor F(1,44)= 0.9715; p= 0.3297,drug factor F(1,44)= 3.094; p= 0.0855], followed by the Turkey’spost hoc test. h Assessment of the hunger level in control and Smarca3cKO group. i Schematic illustration to confirm mossy cell-specific Crerecombination in the hippocampus of MC-Cre transgenic mice. j Co-localization of Cre-dependent reporter (tdTomato, red) with CRT(green, filled arrowheads), but not with PV(green, open arrowheads).Scale bar, 25 µm. k Schematic illustration of genetic deletion ofSmarca3 gene in mossy cells using MC-Cre line. l The NSF test.Control and Smarca3 cKO group have genotype of Smarca3(f/f)

without Cre, and Smarca3(f/f) with MC-Cre, respectively. Two-wayANOVA, [genotype x drug interaction F(1,31)= 4.334; p= 0.0489,genotype factor F(1,31)= 4.404; p= 0.00441, drug factor F(1,31)=2.77; p= 0.3238], followed by the Turkey’s post hoc test. m Thehunger level was assessed by home cage feeding test with the same setof animals. Data are represented as means ± SEM. Pair-wise compar-ison by post hoc test; *p < 0.5, **p < 0.01, ***p < 0.001, ns, non-significant. GCL, granule cell layer; IML, inner molecular layer, CRT,calretinin; PV, parvalbumin; DG, dentate gyrus; MC, mossy cells;SAL, saline; FLX, fluoxetine. See also Figure S1, and S2

Hippocampal mossy cell involvement in behavioral and neurogenic responses to chronic antidepressant. . . 1219

(Fig. 2k). Neither fluoxetine treatment nor inhibition of thep11/AnxA2/SMARCA3 complex caused any significanteffect on home cage feeding (Fig. 2l), suggesting that thisbehavioral change was unlikely due to different hungerlevels between comparison groups. The antidepressant-

induced behavioral change observed in the NSF test wasalso observed in the TST (Figure S5e). Taken together,these data indicate that the p11/AnxA2/SMARCA3 com-plex in mossy cells may play a crucial role in behavioralchanges after chronic treatment with SSRIs.

1220 S.-J. Oh et al.

Effects of cell type-specific inhibition of the p11/AnxA2/SMARCA3 complex on neuronal activity ofmossy cells

We first examined the effects of chronic antidepressantadministration on neuronal activity of mossy cells. Mossycells are spontaneous firing neurons in the hilus, whoseactivity modulates neuronal circuitry and function of thedentate gyrus [46]. To identify mossy cells for electro-physiological recordings, we labeled mossy cells with afluorescence protein, AcGFP1, by injecting Cre-dependentAcGFP1 AAVs into mossy cell-specific Cre (MC-Cre)mice (Fig. 3a). AcGFP1 was shown to be restrictivelyexpressed in mossy cells in the dentate gyrus (Fig. 3b).

Mossy cells were further confirmed by their morphology,tonic firing pattern and membrane capacity (50–100 pF).Electrophysiological recordings showed that chronicfluoxetine treatment significantly enhanced the sponta-neous firing rate of mossy cells (Fig. 3c, d), but acutetreatment of the drug had no effect on mossy cell activity(Fig. 3e, f).

Previous studies have shown that acute restraint stressdecreases c-fos immunoreactivity in hilar mossy cells of thehippocampus [47]. We further investigated whether neuro-nal activity of mossy cells was altered by chronic stressexposure and the following antidepressant administration.After using the chronic unpredictable mild stress (CUMS)paradigm, which is a rodent model of chronic stress-induceddepression [48] (Figure S6a), we counted the active mossycells in the hilus by immunostaining c-Fos, a neural acti-vation marker [47, 49]. c-Fos immunoreactivity in mossycells was significantly reduced by chronic exposure tounpredictable stress (Figures S6b and S6c). In constrast,chronic fluoxetine administration to the stressed animalsresulted in a remarkable increase in c-Fos immunoreactivityin mossy cells (Figures S6d, S6e and S6f), which wasconsistent with our electrophysiological recording shownabove (Fig. 3c, d). These results revealed that the neuronalactivity of mossy cells was suppressed by chronic stress,and the effect was reversible by chronic fluoxetineadministration.

We next examined the regulation by p11 and SMARCA3of mossy cell activity. We found that the spontaneous firingrate of mossy cells was greatly reduced in p11 cKO mice(Fig. 3g, h) and Smarca3 cKO mice (Fig. 3i, j) compared tocontrol mice. In addition, we investigated the effects ofinhibition of the p11/AnxA2/SMARCA3 complex on neu-ronal activity of mossy cells. Previous studies have shownthat the p11/AnxA2/SMARCA3 complex is induced in themossy cells by chronic fluoxetine administration and isfurther targeted to the nuclear matrix to regulate genetranscription [50, 51]. Thus, we additionally generated aPASIP-AcGFP1-NLS construct that was specifically tar-geted to the nucleus of mossy cells due to triple nuclearlocalization signals (3× NLS) fused at the C-terminus(Fig. 3k). We generated a series of Cre-dependent viralvectors, including control AcGFP1, PASIP-AcGFP1, andPASIP-AcGFP1-NLS (Figure S7a). By co-transfection intocultured HEK 293 cells or stereoinjection into the dentategyrus of mossy cell-specific Cre (MC-Cre) transgenic mice,we verified the nuclear-specific expression of the PASIP-AcGFP1-NLS construct (Figure S7b) and also the cell-typespecific expression of the PASIP-AcGFP1 constructs in thedentate mossy cells (Fig. 3l). Of note, the firing rate ofmossy cells was significantly decreased by PASIP-AcGFP1or PASIP-AcGFP1-NLS expression compared to controlAcGFP1 expression (Fig. 3m, n). We further observed that

Fig. 2 Mossy cell-specific disruption of the p11/AnxA2/SMARCA3complex silences neurogenic and behavioral responses to chronicSSRI treatment. a Molecular design for the p11/AnxA2/SMARCA3complex-Inhibitory peptide (PASIP)-AcGFP1 fusion protein (PASIP-AcGFP1). b Schematic illustration of how the recombinant inhibitor(PASIP-AcGFP1) blocks the functional assembly of the p11/AnxA2/SMARCA3 heterohexameric complex. Upon dissociation from thecomplex, SMARCA3 becomes inactive. c In vitro pull-down assaydemonstrates that PASIP-AcGFP1 disrupts the p11/AnxA2/SMARCA3 complex. SMARCA3 was immunoprecipitated from HEK293 cells that were transfected with either control AcGFP1 or PASIP-AcGFP1 construct. The immune complexes were subjected toimmunoblot analysis using AcGFP1-, SMARCA3-, AnxA2- and p11antibodies. d Recombinant AAVs, expressing either control AcGFP1or the PASIP-AcGFP1, were stereotaxically injected into the dentategyrus of MC-Cre transgenic mice. e Representative immunolabelingimages to show mossy cell-specific expression of control AcGFP1 orPASIP-AcGFP1 construct. AcGFP1+ cells (green) were colocalizedwith most CRT+ cells in the hilus (red) as indicated (solid arrow-heads), but not with a few CRT+ immature neurons in the subgranularzone (open arrowheads) to show the cell-type specific gene delivery.Scale bar, 50 μm. f Experimental design. Control AcGFP1-injected orPASIP-AcGFP1-injected mice were administered saline (SAL) orfluoxetine (FLX) for 3 weeks. After behavioral testing, all the animalswere labeled with BrdU for the last 3 h prior to perfusion. g Repre-sentative images with α-BrdU immunostaining results. BrdU+ cells inthe subgranular zone (SGZ) are as indicated in each image (solidarrowheads). Scale bars, 100 μm. h Quantification of BrdU-positivecells in the SGZ. Two-way ANOVA, [AAV × drug interaction F(1,51)= 4.129; p= 0.0498, AAV factor F(1,51)= 18.54; p < 0.0001,drug factor F(1,51)= 8.65; p= 0.0049], followed by the Turkey’s posthoc test. i Representative images of DCX-positive cells in control-AcGFP and PASIP-AcGFP1 mice. j Quantitation of DCX-positivecells in the SGZ and the GCL. Two-way ANOVA, [AAV × druginteraction F(1,25)= 6.364; p= 0.0184, AAV factor F(1,25)= 18.54;p < 0.0001, drug factor F(1,25)= 8.65; p= 0.0049], followed by theTurkey’s post hoc test. k Novelty suppressed feeding test (NSF). Two-way ANOVA, [AAV × drug interaction F(1,54)= 4.09; p= 0.0481,AAV factor F(1,54)= 4.084; p= 0.0327, drug factor F(1,54)= 4.584;p= 0.0368], followed by the Turkey’s post hoc test. l Home cagefeeding levels. Data are represented as means ± SEM. Pair-wisecomparison by post hoc test; *p < 0.5, **p < 0.01, ***p < 0.001, ns,nonsignificant. MC, mossy cells; CRT, calretinin; GCL, granule celllayer; IML, inner molecular layer; DCX, double-cortin; Rel. DCX IF,relative double-cortin immune-fluroscence. See also Figure S3, Fig-ure S4, and Figure S5

Hippocampal mossy cell involvement in behavioral and neurogenic responses to chronic antidepressant. . . 1221

c-Fos immunoreactivity was markedly reduced by PASIP-AcGFP1 expression in mossy cells, but not by controlAcGFP1 expression (Figures S8a–S8c), verifying a sig-nificant role of the p11/AnxA2/SMARCA3 complex in theregulation of mossy cell excitability.

Collectively, these results show that the neural activity ofmossy cells is decreased during chronic stress-induceddepression and this decrease is reversed by chronic anti-depressant administration, possibly through the p11/AnxA2/SMARCA3, an inducible protein complex.

Effects of selective stimulation of dentate mossycells on adult neurogenesis in the hippocampus

Next we examined whether acute chemogenetic stimulation ofmossy cells using the Gq-DREADD system was able to mimicthe chronic effects of fluoxetine. To modulate the dentatemossy cells, we delivered viral vectors expressing controlmCherry (Control) or hM3D-Gq-mCherry (Gq-DREADD),into hilus regions of MC-Cre transgenic mice (Fig. 4a, Fig-ures S9a-S9c). By using immunostaining with c-Fos, we

1222 S.-J. Oh et al.

verified that mossy cells were stimulated by clozapine-N-oxide(CNO) administration in Gq-DREADD mice, when comparedto control mice (Fig. 4b, Figures S9d and S9e).

We assessed adult neurogenesis activities using BrdU-labeling of neural progenitor cells and doublecortin (DCX)-immunolabeling of post-mitotic immature neurons, in thepresence or the absence of CNO/Gq-DREADD activation.Interestingly, Gq-DREADD stimulation of mossy cellssignificantly increased the number of BrdU+ proliferatingcells and DCX immunostaining intensity in the SGZ of thedentate gyrus compared to control mice (Fig. 4c–g). Mossycell-dependent regulation of neurogenesis activities wasconsistent throughout the rostro-caudal axis of the dentategyrus. (Figures S10a-S10d). Next, we investigated the

effects of direct stimulation of mossy cells on basal loco-motion, anxiety, and depression-related behaviors (Fig-ures S11a-g). Chemogenetic-stimulation of mossy cellsalone increased the duration in the open arms of the EPMtest (Figure S11d), but failed to induce significant differ-ences in other types of anxiety-related behaviors (Fig-ures S11a-S11c). Furthermore, acute stimulation of mossycells with CNO injection did not result in any significantchange in the depression-like behaviors including the NSFtest, and the TST (Figures S11e–S11g). Collectively, theseresults suggest that acute stimulation of mossy cells issufficient to trigger adult neurogenesis, but that is unlikelysufficient to achieve full improvement in anxiety anddepressive-like behaviors.

Effect of selective inhibition of dentate mossy cellson antidepressant actions in the hippocampus

Next we examined whether inhibition of the mossy cellsinfluences neurogenic and behavioral effects of chronic anti-depressant treatment. With cre-dependent AAV injections,control mCherry (control) and hM4D-Gi-mCherry (Gi-DREADD) were specifically expressed in the mossy cellsalong the dorso-ventral axis of the dentate gyrus (Fig. 5a, b,Figures S12a–S12c). By using electrophysiological recordings,we verified that the activity of mossy cells was silenced byCNO only in mossy cells expressing Gi-DREADD (Fig. 5c, d).

Using the Gi-DREADD system, we examined the role ofmossy cells on chronic fluoxetine-induced neurogenesis andbehavioral changes (Fig. 5e). We measured adult neuro-genesis activities in control and Gi-DREADD mice.Chronic treatment with SSRI significantly increased thenumber of BrdU+ proliferating cells in the SGZ of thedentate gyrus of control mice, but the fluoxetine effect wassignificantly reduced in the Gi-DREADD mice (Fig. 5f, g,Figures S13a and S13b). We next analyzed the expressionlevel of doublecortin (DCX), a marker for postmitoticimmature neurons. DCX immunostaining intensity wasincreased by chronic fluoxetine treatment in control mice,but to a much lower extent in Gi-DREADD mice (Fig. 5h, i,Figures S13c and S13d). Chronic fluoxetine administrationpromotes the newborn neural progenitors to survive anddifferentiate into mature neurons [52]. Inhibition of mossycell activity results in significant reduction in survival and/or differentiation of post-mitotic newborn cells (Fig. 5h, i).These results indicate that mossy cells may modulatechronic fluoxetine-induced proliferation of neural progeni-tors, differentiation and maturation into new-born granulecells. This phenotype is similar to that observed withgenetic KO or inhibition of the p11/AnxA2/SMARCA3complex in mossy cells.

We further investigated the functional significance ofmossy cell activity by profiling general locomotor activity

Fig. 3 Neuronal activity of mossy cells is enhanced by chronic SSRItreatment via p11/AnxA2/SMARCA3 complex. a Schematic ofAcGFP1 AAV injection into the hippocampus of MC-Cre mice.Ventral hippocampal slices were used for electrophysiological record-ings. b Representative fluorescence images of AcGFP1-labeled mossycells in a hippocampal section of the AAV-injected mice. Magnifica-tions: 5× or 40× objective lens. c, d Representative traces (c) andWhisker box plot (d) showing the spontaneous firing rate of mossycells by chronic FLX administration on 8-week-old mice (oraladministration, 18 days, n= 9 cells/3 mice per group). Scale bar: 1 s,50 mV. Two-tailed, unpaired T-test; p= 0.044. e, f Representativetraces (e) and Whisker box plot (f) showing the spontaneous firing rateof mossy cells from 8-week-old mice injected with a single dose ofsaline or fluoxetine (10mg/kg). n= 9 cells/3 mice per group. Scale bar:1 s, 50 mV. Two-tailed, unpaired T-test; p= 0.9401. g, h Reducedneuronal activity of mossy cells in p11 cKO mice. Mossy cells werelabeled by injecting Cre-dependent AcGFP1 AAV into p11 wild-type(+/+, MC-Cre)(Control) and p11 conditional KO (p11(f/f);MC-Cre) (p11cKO). Representative traces (g) and Whisker box plot (h) showing thespontaneous firing rate of mossy cells in the dentate gyrus of controland p11 cKO mice at 16 weeks of age. n= 10 cells/3 mice for control;n= 6 cells/3 mice for p11 cKO. Scale bar: 300ms, 50mV. Two-tailed,unpaired T-test; p= 0.0255. I, j Reduced electrophysiological activityof mossy cells in Smarca3 cKO mice. Mossy cells were labeled byinjecting the Cre-dependent AcGFP1 AAV into Smarca3 wild-type(+/+, MC-Cre) (Control) and Smarca3 conditional KO (Smarca3(f/f);MC-Cre) (Smarca3 cKO). Representative traces (i) and Whisker box plot(j) showing the spontaneous firing rate of mossy cells in the dentategyrus of control and Smarca3 cKO mice at 16 weeks of age. n= 9cells/3 mice per group. Scale bar: 300 ms, 50 mV. Two-tailed, unpairedT-test; p= 0.0073. k Schematic design for a series of PASIP-fusionconstructs, including control AcGFP1, whole cell inhibitor construct(PASIP-AcGFP1) and nucleus-targeted inhibitor construct (PASIP-AcGFP1-NLS). l Immunofluorescence images showing mossy cell-specific expression in the dentate gyrus of AAV-injected mice.AcGFP1-positive MC was indicated (filled arrowheads). Scale bar, 100μm. m, n Disrupted p11/AnxA2/SMARCA3 complex reduces firingrate of mossy cells. Representative traces (m) and Whisker box plot (n)showing the spontaneous firing rate of mossy cells expressing eachconstruct: control AcGFP1 (n= 12 cells/4 mice), PASIP-AcGFP1 (n=8 cells/3 mice), PASIP-AcGFP1-NLS (n= 8 cells/3 mice) in 16-week-old mice. Scale bar: 200ms, 50 mV. One-way ANOVA [F(2,21)=10.08; p= 0.0009], followed by Bonferroni’s multiple comparison test.Pair-wise comparison; *p < 0.01, **p < 0.01, ***p < 0.001, ns, non-significant; GCL, granule cell layer; IML, inner molecular layer; NLS,nuclear localization signal; MC, mossy cell; SAL, saline; FLX, fluox-etine. See also Figure S6, Figure S7, and Figure S8

Hippocampal mossy cell involvement in behavioral and neurogenic responses to chronic antidepressant. . . 1223

and affective behaviors. Control and Gi-DREADD micedid not display a baseline difference in locomotor activity(open field test [OF], Figure S14a), anxiety-related beha-viors (thigmotaxis, Figure S14b; light/dark box test [LDB],Figure S14c; elevated plus maze test [EPM], Figure S14d).The novelty-suppressed feeding [NSF] test is based ondepressive-like and anxiety behavior and has been widelyused to assess the chronic effect of fluoxetine [53]. Weexamined the possible effect of mossy cell inhibition onbehavioral changes after chronic antidepressant treatment,using the NSF test and the tail suspension test [TST], aclassical depression test. We found that behavioralresponses to chronic fluoxetine treatment were affected byCNO-induced silencing of mossy cell activity. In the NSFtest, the reduced latency to feed after chronic fluoxetinetreatment was abolished in the Gi-DREADD mice(Fig. 5j). In contrast, the home cage feeding level wasunaffected, which suggests that this behavioral change is

unlikely due to different hunger levels between compar-ison groups (Fig. 5k). In the TST, the effect of fluoxetineon immobility was significantly reduced in controlmice, but not in Gi-DREADD mice (Figure S14e). Col-lectively, our results demonstrate that modulation of mossycell activity in the dentate gyrus is critical for both neu-rogenic and behavioral responses to chronic fluoxetineadministration.

Discussion

In the present study, we have demonstrated a role forhippocampal mossy cells in mediating the behavioraland neurogenic responses to chronic antidepressanttreatment. Furthermore, we have shown that p11 andSMARCA3 play crucial roles in modulating mossy cellactivity (Figure S15).

Fig. 4 Acute chemogenetic stimulation of mossy cells increases theadult neurogenesis activities in the dentate gyrus. a Schematic illus-tration of chemogenetic stimulation of mossy cells with Gq-DREADDsystem. b Representative images of c-Fos positive cells (Green) in thedentate hilus of the control or Gq-DREADD mice, 2 h after CNOinjection. Scale bar, 50 μm. c Schematic of experimental design.Control and Gq-DREADD mice were tested for anxiety-related ordepression-like behaviors with CNO injection. One day after the lastbehavioral testing, all mice were labeled with BrdU for 3 h and thensacrificed with transcardinal perfusion. d Representative images of

BrdU immunostaining in the dorsal hippocampal sections from thecontrol or Gq-DREADD mice. e Quantitation of BrdU-positive cells inthe SGZ. Two-tailed, unpaired T-test; p < 0.0001. f Representativeimages of DCX-immunostaining in the dorsal hippocampal sectionsfrom the control and Gq-DREADD mice. g Quantitation of relativeintensity of DCX immunofluroscence in the SGZ and the GCL of thedentate gyrus. Data are represented as means ± SEM. Two-tailed,unpaired T-test; p= 0.0021. CNO, clozapine-N-oxide; SGZ, sub-granular zone; GCL, granule cell layer; IML, inner molecular layer.See also Figure S9, Figure S10 and Figure S11

1224 S.-J. Oh et al.

Role of mossy cells in the behavioral response tochronic SSRI medication

In addition to established roles in cognitive processes, thehippocampus has been associated with emotional control bytaking a prominent position in the limbic circuitry. There is

a well-established link between hippocampal dysfunctionand the pathogenesis of mood-related disorders. In addition,antidepressive medication has been shown to enhanceneuronal plasticity in the hippocampus [54–56]. It thus hasbeen regarded as an important target area for the develop-ment of advanced antidepressant therapeutics [3, 57].

Hippocampal mossy cell involvement in behavioral and neurogenic responses to chronic antidepressant. . . 1225

Mossy cells are very active as reflected by spontaneousaction potentials and high-frequency spontaneous excitatorypostsynaptic currents (EPSCs) [10, 58]. Here, we found thatchronic exposure to unpredictable stress suppresses mossycell activity as measured by c-fos expression, a neuronalexcitation marker, which is reversed by chronic SSRIadministration. In addition, our electrophysiological record-ings showed that the tonic firing rate of mossy cells isenhanced by chronic, but not acute, treatment with an SSRI,fluoxetine, which is consistent with the induction pattern ofp11 and its ternary complex p11/AnxA2/SMARCA3 in thehippocampus [34]. In contrast, mossy cell-specific knockoutof p11 or of Smarca3 or a disruption of the p11/AnxA2/SMARCA3 complex reduces the tonic firing and c-fosexpression of mossy cells, suggesting a significant role of theternary complex in regulating the neuronal activity of mossycells. These results suggest that mossy cells are labile neuronsthat undergo considerable neuroplastic changes in response toexternal stress or chronic antidepressant administration.

Despite the potential involvement of mossy cells in neu-rological and neuropsychiatric disorders, our understanding

of their functions and pathological implications remainslimited. To date, few studies have been performed toexamine animal behaviors by manipulating mossy cells. Aprevious study deleted mossy cells by using a cytotoxic toxinexpression and found transient elevation of anxiety andimpaired pattern separation in rodents [23]. A recent studyadopted an optogenetic approach and showed that mossycells control spontaneous convulsive seizures and spatialmemory [59]. Here we demonstrate that behavioral effects ofchronic fluoxetine treatment are significantly diminishedwhen mossy cell activity is suppressed. Furthermore, che-mogenetic inhibition of mossy cell activity impairs the SSRI-induced changes in affective behaviors. Collectively, ourfindings provide strong evidence that mossy cells play asignificant role in antidepressant actions, in addition to theirreported role in cognitive behaviors. However, acute che-mogenetic stimulation of mossy cells is unlikely to be suf-ficient to achieve full behavioral improvement as seen bychronic SSRI administration. Behavioral benefit after chronicSSRI administration may require cooperative neuroplasticchanges in multiple neural cell types that include hippo-campal mossy cells.

Role of mossy cells in the neurogenic response tochronic SSRI medication

It has been unknown whether mossy cells are involved inadult neurogenesis, although a previous study suggestedthat mossy cells provide the first excitatory input to adult-born granule cells [60]. The current study is the first to showthat mossy cells influence SSRI-induced neurogenesis. Wereport here that mossy cells, regulated by the p11/AnxA2/SMARCA3 complex, mediate the neurogenic responses tochronic antidepressant administration. Consistent with ourprevious reports using p11 or Smarca3 conventional KOmice [29, 34], disruption of the p11/AnxA2/SMARCA3complex in mossy cells impairs adult neurogenesis inresponse to chronic fluoxetine treatment. Consistently,chemogenetic inhibition of mossy cells results in a decreasein SSRI-induced adult neurogenesis in the hippocampus.Furthermore, acute chemogenetic stimulation of mossy cellsis able to trigger early neurogenic activities including theproliferation of the neural stem cells, survival, and differ-entiation into new-born granule cells. Thus, these resultssuggest crucial roles of mossy cells in multiple stages ofSSRI-induced neurogenesis in the hippocampus.

Despite extensive studies performed over the past 20years, the molecular mechanisms underlying SSRI-inducedneurogenesis remain unclear. Several genetic and pharma-cological approaches have demonstrated that 5-HT1A and5-HT4 receptors, enriched in dentate granule cells, arecrucial for SSRI-induced neurogenesis [61–63]. In contrast,the basal maintenance level of hippocampal neurogenesis is

Fig. 5 Chemogenetic silencing of mossy cells attenuates neurogenicand behavioral responses to chronic SSRIs. a Schematic illustration ofchemogenetic silencing of mossy cells with Gi-DREADD system. bRepresentative images for mossy cell-specific delivery of Gi-DREADD system. Hippocampal sections from AAV-injected micewere co-labeled with a mossy cell marker, CRT. Scale bar, 50 μm. c, dRepresentative traces (c) and dot plot (d) showing the spontaneousfiring of mossy cells expressing either control mCherry (Control) orGi-DREADD-mCherry (Gi-DREADD) before and after bath applica-tion of CNO (1 μM). n= 4 cells per group. Scale bar: 10 s, 50 mV.Two-way ANOVA with repeated measurement, [AAV x CNO inter-action F(1,6)= 7.261; p= 0.0395, CNO factor F(1,6)= 6.473; p=0.0438, CNO factor F(1,6)= 9.709; p= 0.0321], followed by theSidak’s multiple comparison post hoc test. Control (CNO effect); p=0.9355, Gi-DREADD (CNO effect); *p= 0.0343. e Schematic ofexperimental design. Control and Gi-DREADD mice were admini-strated saline (SAL) or fluoxetine (FLX) for 3 weeks and labeled withBrdU for the last 3 h prior to perfusion. f Representative images tovisualize proliferating neural stem cells with α-BrdU immunostaining(BrdU, Green, filled arrowheads). g BrdU+ cells were quantified alongthe subgranular zone. Two-way ANOVA, [AAV × drug interactionF(1,28)= 3.548; p= 0.0700, AAV F(1,28)= 1.707; p= 0.2020, drugF(1,28)= 13.55; p= 0.001], followed by the Turkey’s post hoc test.h Postmitotic immature neurons were immunostained withα-doublecortin antibody (DCX, green). Scale bar, 100 μm. i Quanti-fication of relaltive DCX immunofluorescence intensity (Rel. DCX IF)were quantified in the sub-granular and granular zone using ImageJ software. Pair-wise comparison with two-tailed, unpaired T-test.j Effect of Gi-DREADD-dependent inhibition of mossy cells on thenovelty-suppressed feeding test after chronic fluoxetine treatment.Two-way ANOVA, [AAV x drug interaction F(1,28)= 4.212; p=0.0496, AAV factor F(1,28)= 1.39; p= 0.2484, drug factor F(1,28)=15.11; p= 0.0006], followed by the Turkey’s post hoc test. k Homecage feeding test. Data are represented as means ± SEM. Pair-wisecomparison; *p < 0.5, **p < 0.01, ***p < 0.001. ns, nonsignificant;CRT, calretinin; CNO, clozapine-N-oxide; GCL, granule cell layer;IML, inner molecular layer. See also Figure S12, Figure S13, andFigure S14

1226 S.-J. Oh et al.

not altered in either 5-HT1A receptor-deficient or 5-HT4receptor-deficient mice, which indicates the specific invol-vement of these serotonin pathways in SSRI-induced neu-rogenesis. Here we observed that the p11/AnxA2/SMARCA3 complex in mossy cells regulates SSRI-inducedneurogenesis, but not the basal level. Given that mossy cellsand granule cells make extensive reciprocal connections toconstruct the neural circuitry of the dentate gyrus, it isplausible that serotonergic regulation of both granule cellsand mossy cells may cooperate to induce the neurogenesisand further behavioral response to chronic SSRIadministration.

In conclusion, we have demonstrated that mossy cellsregulate affective behaviors as well as neurogenesis inresponse to SSRIs, in addition to their established role incognitive behaviors. The present study contributes to a betterunderstanding of the neural mechanisms mediating the ben-eficial effects of SSRI medication and should contribute tothe development of novel antidepressant therapeutics to targetspecific neural cell types in the mood-regulatory circuits.

Acknowledgements We are grateful to Dr. Helen Scharfman (NewYork University, USA) for helpful advice and discussion and also toDr. Kazu Nakazawa (University of Alabama, USA) for kindly sharing[Calcrl]-Cre transgenic mice. This research was supported by theBrain Research Program through the National Research Foundation ofKorea (NRF) funded by the Ministry of Science, ICT (NRF-2017M3C7A1048448 to YSO); the Basic Science Research Programthrough the National Research Foundation of Korea (NRF) funded bythe Ministry of Education (NRF-2016R1D1A1B03935615 to YSO);the Bio & Medical Technology Development Program (NRF-2017M3A9G8084463 to YSO); the DGIST R&D Program of theMinistry of Science, ICT and Future Planning (18-BD-0402 to YSO);KBRI basic research program through Korea Brain Research Institutefunded by Ministry of Science and ICT (18-BR-04-03 to YSO); 2014NARSAD YI AWARD (Grant No. 20695 to YSO). In addition, thiswork was supported by the United States Army Medical Research andMaterial Command (USAMRMC) under Award No.W81XWH-16-1-0681 (to PG), funds received from The JPB Foundation, Award No.475 (to PG) and funds received from the Black Family Foundation(to PG).

Author contributions SO, JC, PG, and YSO designed the experiments.SO and JA performed and analyzed the behavior test. JC performed andanalyzed electrophysiology experiments. SO performed and analyzedimmunofluorescence data. SO, JC, PG, and YSO wrote the paper.

Compliance with ethical standards

Conflict of interest The authors declare that they have no conflict ofinterest.

Publisher’s note: Springer Nature remains neutral with regard tojurisdictional claims in published maps and institutional affiliations.

References

1. Duman RS, Aghajanian GK. Synaptic dysfunction in depression:potential therapeutic targets. Science. 2012;338:68–72.

2. Nestler EJ, Barrot M, DiLeone RJ, Eisch AJ, Gold SJ, MonteggiaLM. Neurobiology of depression. Neuron. 2002;34:13–25.

3. Berton O, Nestler EJ. New approaches to antidepressant drug dis-covery: beyond monoamines. Nat Rev Neurosci. 2006;7:137–51.

4. Holtzheimer PE, Mayberg HS. Stuck in a rut: rethinking depres-sion and its treatment. Trends Neurosci. 2011;34:1–9.

5. Krishnan V, Nestler EJ. Linking molecules to mood: new insightinto the biology of depression. Am J Psychiatry. 2010;167:1305–20.

6. Covington HE 3rd, Vialou V, Nestler EJ. From synapse tonucleus: novel targets for treating depression. Neuropharmacol-ogy. 2010;58:683–93.

7. Duman RS, Aghajanian GK, Sanacora G, Krystal JH. Synapticplasticity and depression: new insights from stress and rapid-acting antidepressants. Nat Med. 2016;22:238–49.

8. Rush AJ, Trivedi MH, Wisniewski SR, Nierenberg AA, StewartJW, Warden D, et al. Acute and longer-term outcomes indepressed outpatients requiring one or several treatment steps: aSTAR*D report. Am J Psychiat. 2006;163:1905–17.

9. Amaral DG. A Golgi study of cell types in the hilar region of thehippocampus in the rat. J Comp Neurol. 1978;182:851–914.

10. Scharfman HE. Electrophysiological evidence that dentate hilarmossy cells are excitatory and innervate both granule cells andinterneurons. J Neurophysiol. 1995;74:179–94.

11. Blasco-Ibanez JM, Freund TF. Distribution, ultrastructure, andconnectivity of calretinin-immunoreactive mossy cells of themouse dentate gyrus. Hippocampus. 1997;7:307–20.

12. Henze DA, Buzsaki G. Hilar mossy cells: functional identificationand activity in vivo. Prog Brain Res. 2007;163:199–216.

13. Scharfman HE. The enigmatic mossy cell of the dentate gyrus. NatRev Neurosci. 2016;17:562–75.

14. Jinde S, Zsiros V, Nakazawa K. Hilar mossy cell circuitry con-trolling dentate granule cell excitability. Front Neural Circuits.2013;7:14.

15. Sun Y, Grieco SF, Holmes TC, Xu X. Local and long-range cir-cuit connections to hilar mossy cells in the dentate gyrus. eNeuro.2017;4. https://doi.org/10.1523/ENEURO.0097-17.2017.

16. Gage FH, Thompson RG. Differential distribution of nor-epinephrine and serotonin along the dorsal-ventral axis of thehippocampal formation. Brain Res Bull. 1980;5:771–3.

17. Lisman JE, Grace AA. The hippocampal-VTA loop: controllingthe entry of information into long-term memory. Neuron. 2005;46:703–13.

18. Patel A, Bulloch K. Type II glucocorticoid receptor immunor-eactivity in the mossy cells of the rat and the mouse hippocampus.Hippocampus. 2003;13:59–66.

19. Danielson NB, Turi GF, Ladow M, Chavlis S, Petrantonakis PC,Poirazi P, et al. In vivo imaging of dentate gyrus mossy cells inbehaving mice. Neuron. 2017;93:552–59.

20. GoodSmith D, Chen X, Wang C, Kim SH, Song H, Burgalossi A,et al. Spatial representations of granule cells and mossy cells of thedentate gyrus. Neuron. 2017;93:677–90.

21. Senzai Y, Buzsaki G. Physiological properties and behavioralcorrelates of hippocampal granule cells and mossy cells. Neuron.2017;93:691–704.

22. Nakazawa K. Dentate mossy cell and pattern separation. Neuron.2017;93:465–7.

23. Jinde S, Zsiros V, Jiang Z, Nakao K, Pickel J, Kohno K, et al.Hilar mossy cell degeneration causes transient dentate granule cellhyperexcitability and impaired pattern separation. Neuron.2012;76:1189–200.

24. Svenningsson P, Chergui K, Rachleff I, Flajolet M, Zhang X, ElYacoubi M, et al. Alterations in 5-HT1B receptor function by p11in depression-like states. Science. 2006;311:77–80.

25. Svenningsson P, Kim Y, Warner-Schmidt J, Oh YS, Greengard P.p11 and its role in depression and therapeutic responses to anti-depressants. Nat Rev Neurosci. 2013;14:673–80.

Hippocampal mossy cell involvement in behavioral and neurogenic responses to chronic antidepressant. . . 1227

26. Alexander B, Warner-Schmidt J, Eriksson T, Tamminga C,Arango-Lievano M, Ghose S, et al. Reversal of depressed beha-viors in mice by p11 gene therapy in the nucleus accumbens. SciTransl Med. 2010;2:54ra76.

27. Anisman H, Du L, Palkovits M, Faludi G, Kovacs GG, Szontagh-Kishazi P, et al. Serotonin receptor subtype and p11 mRNAexpression in stress-relevant brain regions of suicide and controlsubjects. J Psychiatry Neurosci. 2008;33:131–41.

28. Alexander B, Warner-Schmidt J, Eriksson T, Tamminga C,Arango-Lievano M, Ghose S, et al. Reversal of depressed beha-viors in mice by p11 gene therapy in the nucleus accumbens. SciTransl Med. 2010;2:54ra76.

29. Egeland M, Warner-Schmidt J, Greengard P, Svenningsson P.Neurogenic effects of fluoxetine are attenuated inp11 (S100A10)knockout mice. Biol Psychiatry. 2010;67:1048–56.

30. Warner-Schmidt JL, Chen EY, Zhang X, Marshall JJ, Morozov A,Svenningsson P, et al. A role for p11 in the antidepressant actionof brain-derived neurotrophic factor. Biol Psychiatry.2010;68:528–35.

31. Schmidt EF, Warner-Schmidt JL, Otopalik BG, Pickett SB,Greengard P, Heintz N. Identification of the cortical neurons thatmediate antidepressant responses. Cell. 2012;149:1152–63.

32. Warner-Schmidt JL, Schmidt EF, Marshall JJ, Rubin AJ, Arango-Lievano M, Kaplitt MG, et al. Cholinergic interneurons in thenucleus accumbens regulate depression-like behavior. Proc NatlAcad Sci USA. 2012;109:11360–5.

33. Eriksson TM, Alvarsson A, Stan TL, Zhang X, Hascup KN,Hascup ER, et al. Bidirectional regulation of emotional memoryby 5-HT1B receptors involves hippocampalp11. Mol Psychiatry.2013;18:1096–105.

34. Oh YS, Gao P, Lee KW, Ceglia I, Seo JS, Zhang X, et al.SMARCA3, a chromatin-remodeling factor, is required for p11-dependent antidepressant action. Cell. 2013;152:831–43.

35. Lee KW, Westin L, Kim J, Chang JC, Oh YS, Amreen B, et al.Alteration by p11 of mGluR5 localization regulates depression-like behaviors. Mol Psychiatry. 2015;20:1546–56.

36. Milosevic A, Liebmann T, Knudsen M, Schintu N, SvenningssonP, Greengard P. Cell- and region-specific expression ofdepression-related proteinp11 (S100a10) in the brain. J CompNeurol. 2017;525:955–75.

37. Seo JS, Zhong P, Liu A, Yan Z, Greengard P. Elevation of p11 inlateral habenula mediates depression-like behavior. Mol Psy-chiatry 2017;23:1113.

38. Gerke V, Creutz CE, Moss SE. Annexins: linking Ca2+ signallingto membrane dynamics. Nat Rev Mol Cell Biol. 2005;6:449–61.

39. Medrihan L, Sagi Y, Inde Z, Krupa O, Daniels C, Peyrache A,et al. Initiation of behavioral response to antidepressants by cho-lecystokinin neurons of the dentate gyrus. Neuron 2017;95:564–76.

40. Gangarossa G, Longueville S, De Bundel D, Perroy J, Herve D,Girault JA, et al. Characterization of dopamine D1 and D2receptor-expressing neurons in the mouse hippocampus. Hippo-campus. 2012;22:2199–207.

41. Puighermanal E, Biever A, Espallergues J, Gangarossa G, DeBundel D, Valjent E. drd2-cre:ribotag mouse line unravels thepossible diversity of dopamine d2 receptor-expressing cells of thedorsal mouse hippocampus. Hippocampus. 2015;25:858–75.

42. Lucassen PJ, Meerlo P, Naylor AS, van Dam AM, Dayer AG,Fuchs E, et al. Regulation of adult neurogenesis by stress, sleepdisruption, exercise and inflammation: implications for depressionand antidepressant action. Eur Neuropsychopharmacol. 2010;20:1–17.

43. Duman RS, Malberg J, Nakagawa S, D’Sa C. Neuronal plasticityand survival in mood disorders. Biol Psychiatry. 2000;48:732–9.

44. Santarelli L, Saxe M, Gross C, Surget A, Battaglia F, Dulawa S,et al. Requirement of hippocampal neurogenesis for the behavioraleffects of antidepressants. Science. 2003;301:805–9.

45. Couillard-Despres S, Winner B, Schaubeck S, Aigner R, VroemenM, Weidner N, et al. Doublecortin expression levels in adult brainreflect neurogenesis. Eur J Neurosci. 2005;21:1–14.

46. Scharfman HE, Myers CE. Hilar mossy cells of the dentate gyrus:a historical perspective. Front Neural Circuits. 2012;6:106.

47. Moretto JN, Duffy AM, Scharfman HE. Acute restraint stressdecreases c-fos immunoreactivity in hilar mossy cells of the adultdentate gyrus. Brain Struct Funct. 2017;222:2405–19.

48. Willner P. Chronic mild stress (CMS) revisited: consistency andbehavioural-neurobiological concordance in the effects of CMS.Neuropsychobiology. 2005;52:90–110.

49. Dragunow M, Robertson HA. Kindling stimulation induces c-fosprotein(s) in granule cells of the rat dentate gyrus. Nature.1987;329:441–2.

50. Das S, Shetty P, Valapala M, Dasgupta S, Gryczynski Z, Vish-wanatha JK. Signal transducer and activator of transcription 6(STAT6) is a novel interactor of annexin A2 in prostate cancercells. Biochemistry. 2010;49:2216–26.

51. Liu J, Vishwanatha JK. Regulation of nucleo-cytoplasmic shut-tling of human annexin A2: a proposed mechanism. Mol CellBiochem. 2007;303:211–20.

52. Wang JW, David DJ, Monckton JE, Battaglia F, Hen R. Chronicfluoxetine stimulates maturation and synaptic plasticity of adult-born hippocampal granule cells. J Neurosci. 2008;28:1374–84.

53. David DJ, Samuels BA, Rainer Q, Wang JW, Marsteller D,Mendez I, et al. Neurogenesis-dependent and -independent effectsof fluoxetine in an animal model of anxiety/depression. Neuron.2009;62:479–93.

54. Karpova NN, Pickenhagen A, Lindholm J, Tiraboschi E,Kulesskaya N, Agustsdottir A, et al. Fear erasure in mice requiressynergy between antidepressant drugs and extinction training.Science. 2011;334:1731–4.

55. Castrén E. Is mood chemistry? Nat Rev Neurosci. 2005;6:241.56. Shuto T, Kuroiwa M, Sotogaku N, Kawahara Y, Oh Y-S, Jang

J-H, et al. Obligatory roles of dopamine D1 receptors in the dentategyrus in antidepressant actions of a selective serotonin reuptakeinhibitor, fluoxetine. Mol Psychiatry. 2018 https://doi.org/10.1038/s41380-018-0316-x. [EPub ahead of print].

57. Krishnan V, Nestler EJ. The molecular neurobiology of depres-sion. Nature. 2008;455:894–902.

58. Scharfman HE, Schwartzkroin PA. Electrophysiology of mor-phologically identified mossy cells of the dentate hilus recorded inguinea pig hippocampal slices. J Neurosci. 1988;8:3812–21.

59. Bui AD, Nguyen TM, Limouse C, Kim HK, Szabo GG, Felong S,et al. Dentate gyrus mossy cells control spontaneous convulsiveseizures and spatial memory. Science. 2018;359:787–90.

60. Chancey JH, Poulsen DJ, Wadiche JI, Overstreet-Wadiche L.Hilar mossy cells provide the first glutamatergic synapses to adult-born dentate granule cells. J Neurosci. 2014;34:2349–54.

61. Imoto Y, Kira T, Sukeno M, Nishitani N, Nagayasu K, NakagawaT, et al. Role of the 5-HT4 receptor in chronic fluoxetinetreatment-induced neurogenic activity and granule cell dema-turation in the dentate gyrus. Mol Brain. 2015;8:29.

62. Samuels BA, Anacker C, Hu A, Levinstein MR, Pickenhagen A,Tsetsenis T, et al. 5-HT1A receptors on mature dentate gyrusgranule cells are critical for the antidepressant response. NatNeurosci. 2015;18:1606–16.

63. Segi-Nishida E. The effect of serotonin-targeting antidepressantson neurogenesis and neuronal maturation of the hippocampusmediated via 5-HT1A and 5-HT4 receptors. Front Cell Neurosci.2017;11:142.

1228 S.-J. Oh et al.